Q Acoustics 3050i Room measurement and correction with Brutefir

[ezublab]

Hello,
This is my first post...
I'd like to share my measurements to get advices...
Loudspeakers are Qacoustics QA3050i
Amp : Old Rega Brio (28 years old!)
Streamer/DAC : Allo Boss + Rpi3
Software : Volumio + Dsp plugin.
Filters calculated with DRC-FIR (from the plugin, using a home made target curve).
The room is about 35m2

First pict : unfiltered. Sweep recorded from listening point :

Second picture : Sweep after correction recorded from listening point :

Left waterfall after correction :

Right waterfall after correction :

Both speakers using pink noise + RTA measure :

What is your opinion?

 

[raindance]

How does it sound? That's a lot of correction! I had a pair of the first series of those and thought they were quite laid back sounding. They could not excite many modes in my room unless I put them quite close to the wall (looks like you have quite a lot of bass, I got very little; my room might be too large). Looks like fixing the crossover bump has cost you some efficiency.

 

[ezublab]

It sounds rather good. Each speaker is 30cm from the back wall and more than 1m from corners of the room.

Looks like fixing the crossover bump has cost you some efficiency.

What do you mean?

 

[raindance]

Level has gone down by a few dB is what I mean. This tends to happen when you remove peaks. So you'd need to turn the volume up a little. No biggie.

 

[QMuse]

Hello,
This is my first post...
I'd like to share my measurements to get advices...
Loudspeakers are Qacoustics QA3050i
Amp : Old Rega Brio (28 years old!)
Streamer/DAC : Allo Boss + Rpi3
Software : Volumio + Dsp plugin.
Filters calculated with DRC-FIR (from the plugin, using a home made target curve).
The room is about 35m2

First pict : unfiltered. Sweep recorded from listening point :View attachment 71038


Second picture : Sweep after correction recorded from listening point :
View attachment 71039

Left waterfall after correction :
View attachment 71040

Right waterfall after correction :
View attachment 71041

Both speakers using pink noise + RTA measure :
View attachment 71043

What is your opinion?

Well done!

 

[shumi]

ezublab, can you do me a big favor please? Can you rerun a sweep with the correction with a target average dB of 78?

 

[tecnogadget]

Why use a 130dB scaling in the Waterfall plots ?? It’s recomendable for correct interpretation to share plots with no more than 50dB (example: from 90dB to 40dB), the exact values will be the ones that lets you center the response or at least put the important stuff in either the superior or inferior third of the graph.

 

[Daverz]

Why use a 130dB scaling in the Waterfall plots ?? It’s recomendable for correct interpretation to share plots with no more than 50dB (example: from 90dB to 40dB), the exact values will be the ones that lets you center the response or at least put the important stuff in either the superior or inferior third of the graph.

Yes, typical y-axis limits are 45 to 105 dB. This gives a consistent scale for reporting response plots.

Are you using one of the DRC-FIR "presets" (e.g. normal, minimal, etc), or have you customized the DRC config?

One issue with DRC-FIR is that it corrects each channel without reference to the other one and doesn't attempt to balance channels. You may need to adjust the channel attenuation levels in your brutefir config to get a centered phantom image. Pink noise is useful for this.

 

[ezublab]

Here some new measures... I try to get 78dB...
I added new waterfall with a reduce scale...
Filters are calculated with DRC-FIR normal.

Left and right

Left waterfall

Right waterfall

 

[tecnogadget]

It looks nice, you get the classic slope down house curve with 10dB from 30Hz to 10kHz in your case.
The slope suddenly rolls of from 10kHz to 20kHz, have you tried getting that same 10dB tilt from 30Hz to 20kHz? And maybe some slight boost of 3dB below 100Hz.

Overall it looks much better than the uncorrected response. What’s your subjective impressions of the Before/After ?

 

[shumi]

Thanks a lot ezublab. I was concerned about room modes not scaling linearly with volume increases as your original traces were generated at failry low volume....but I was wrong. They look great at 78dB. You have me inspired to get it done but I have hit a block though. When I follow the instructions from
https://volumio.github.io/docs/Plugins_User_Manuals/brutefir/Making_filters.html
and save my traces in REW for rePhase equalizer, by using "save filter to file", REW saves an XML file which is not going to be read by rePhase. If I use "export filter setting as texts", rePhase is also not happy with the text format. HELP!

 

[Daverz]

Could you paste in your target curve?

Here's my last DRC-FIR correction using the "normal" settings and the default "psychoacoustic" curve that gives a slight tilt to the in-room response (looks like about 6dB drop between 40 Hz and 10 kHz). This gives a subjectively well-damped bass sound that is actually pretty good for my second floor condo. For more excitement (and annoyance for the neighbors), I'd probably shelve the bass up a bit.

 

[Daverz]

When I follow the instructions from
https://volumio.github.io/docs/Plugins_User_Manuals/brutefir/Making_filters.html
and save my traces in REW for rePhase equalizer, by using "save filter to file", REW saves an XML file which is not going to be read by rePhase. If I use "export filter setting as texts", rePhase is also not happy with the text format. HELP!

Hopefully someone can answer your rePhase question; I know I've successfully loaded the XML output in rePhase before, but haven't tried this in a while.

However, unless you are trying to do some kind of customization in rePhase, you can generate filter impulse response files directly from REW by using File->Export->Export filters impulse response as WAV.

 

[ezublab]

@Daverz I use HK9 provides in the plugin.

 

[Daverz]

@Daverz I use HK9 provides in the plugin.

Note that if you don't turn off "psychoacoustic target" in your drc-fir config, the target levels in your file will get added to the psychoacoustic target curve.

You can turn off the psychoacoustic mode from the command-line or by editing the config file. On the command-line you'd use the option --PTType=N for no psychoacoustic target:

drc --PTType=N --BCInFile=uncorrected_impulse.pcm --PSOutFile=correction_filter.pcm --PSPointsFile=bk48.txt  config.drc

Or you could set

PTType = N

in config.drc.

 

[ezublab]

This is the mode used in the plugin.

 

[Daverz]

This is the mode used in the plugin.

Ah, thanks, I'm not familiar with the plugin, it sounds like they've done a great job making DRC-FIR accessible. I use my own bash scripts that run the drc commands and create brutefir config files for each filter.